Tools
Sorted by:
Try your query at:
 |
 |
 |
 |
 |
 |
Results 1 - 10
of
13,271
Table 2. Pronunciation variability for the 100 most common words in the phonetically segmented portion of the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from a variant of the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [15]. Reprinted from [14].
in Speaking In Shorthand -- A Syllable-Centric Perspective For Understanding Pronunciation Variation
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table2 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Many of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 3: ... One potentially useful representation is at the level of the syllable. The 30 most common words in the Switchboard corpus are monosyllabic ( Table2 ), and of the 100 most frequent lexical items only ten are not (and all of these contain but two syllables). This decided lexical preference for syllabic brevity among the most frequently occurring words is largely representative of the corpus as a whole.... ..."
Table 2. Pronunciation variability for the 100 most common words in the phonetically segmented portion of the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from a variant of the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [15]. Reprinted from [14].
in Speaking In Shorthand -- A Syllable-Centric Perspective For Understanding Pronunciation Variation
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table2 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Many of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 3: ... One potentially useful representation is at the level of the syllable. The 30 most common words in the Switchboard corpus are monosyllabic ( Table2 ), and of the 100 most frequent lexical items only ten are not (and all of these contain but two syllables). This decided lexical preference for syllabic brevity among the most frequently occurring words is largely representative of the corpus as a whole.... ..."
Table 2. Pronunciation variability for the 100 most common words in the phonetically segmented portion of the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from a variant of the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [15]. Reprinted from [14].
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table2 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Many of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 3: ... One potentially useful representation is at the level of the syllable. The 30 most common words in the Switchboard corpus are monosyllabic ( Table2 ), and of the 100 most frequent lexical items only ten are not (and all of these contain but two syllables). This decided lexical preference for syllabic brevity among the most frequently occurring words is largely representative of the corpus as a whole.... ..."
Table 4 shows the overlap between the phonological rule lexicon and two of the data-derived lexica: phone recognition and d-trees, respectively. The number of variants generated by each of the phonological rules is shown in column 2 of Table 4. Combi indicates those variants that are the result of a combination of rules applying to a word. For the phone recognition and the d-trees lexicon, the number of variants for each of the rules was determined by comparing them to the phonological rule lexicon and counting the overlap. In columns 4 and 6, the percentage of variants in the phonological rule lexicon that is covered by the phone recognition and d-trees lexica, respectively, is shown.
2000
"... In PAGE 4: ... Table4 : Number of variants present in the phonological rules lexicon, as a result of phone recognition, and after smoothing phone recognition with d-trees. Percentages indicate the proportion of variants in the phonological rule lexicon that is covered by the other two lexica.... In PAGE 4: ... Percentages indicate the proportion of variants in the phonological rule lexicon that is covered by the other two lexica. Table4 shows us that in total 10% of the variants present in the phonological rule lexicon are also found in the phone recognition lexicon. Using d-trees to smooth the phone recognition leads to 23% overlap between the phonological rule variants and the data-derived variants.... ..."
Cited by 8
Table 1. Pronunciation variability for the 100 most common words in the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [18].
1997
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table1 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Most of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 6: ...range of variability, which is often quite broad. It is not uncommon for frequently occurring words to be phonetically realized in dozens of different ways, with the most popular variant often accounting for only 10-15% of the forms ( Table1 ). However, the patterns of phonetic variation are relatively straightforward to describe within a syllabic framework.... In PAGE 8: ... One of the hallmarks of human discourse is its flexible format and expression. There are many different ways to pronounce common words ( Table1 ) and embed these within a grammatical framework that is itself rather fluidly defined. Despite such variability in linguistic expression, listeners readily interpret the speech stream.... ..."
Cited by 38
Table 1. Pronunciation variability for the 100 most common words in the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [18].
1997
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table1 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Most of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 6: ...range of variability, which is often quite broad. It is not uncommon for frequently occurring words to be phonetically realized in dozens of different ways, with the most popular variant often accounting for only 10-15% of the forms ( Table1 ). However, the patterns of phonetic variation are relatively straightforward to describe within a syllabic framework.... In PAGE 8: ... One of the hallmarks of human discourse is its flexible format and expression. There are many different ways to pronounce common words ( Table1 ) and embed these within a grammatical framework that is itself rather fluidly defined. Despite such variability in linguistic expression, listeners readily interpret the speech stream.... ..."
Cited by 38
Table 1. Pronunciation variability for the 100 most common words in the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [18].
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table1 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Most of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 6: ...range of variability, which is often quite broad. It is not uncommon for frequently occurring words to be phonetically realized in dozens of different ways, with the most popular variant often accounting for only 10-15% of the forms ( Table1 ). However, the patterns of phonetic variation are relatively straightforward to describe within a syllabic framework.... In PAGE 8: ... One of the hallmarks of human discourse is its flexible format and expression. There are many different ways to pronounce common words ( Table1 ) and embed these within a grammatical framework that is itself rather fluidly defined. Despite such variability in linguistic expression, listeners readily interpret the speech stream.... ..."
Table 1. Pronunciation variability for the 100 most common words in the Switchboard Transcription Corpus. quot;N quot; is the number of instances each word appears in the 72-minute corpus. quot;#Pr. quot; is the number of distinct phonetic expressions for each word. quot;%Tot quot; is the percentage of the total number of pronunciations accounted for by the single most common variant. The phonetic representation is derived from the Arpabet orthography. Further details concerning both the pronunciation data and the transcription orthography may be found in [18].
"... In PAGE 3: ... One hundred words account for fully 66% of the individual tokens (Figure 2). A perusal of these most frequently occurring words ( Table1 ) indicates that most come from the so-called quot;closed quot; or quot;function quot; class words such as pronouns, articles, conjunctions and modal/auxiliary verbs. Most of the remainder stem from just a few basic nominal, adjectival or verbal forms.... In PAGE 6: ...range of variability, which is often quite broad. It is not uncommon for frequently occurring words to be phonetically realized in dozens of different ways, with the most popular variant often accounting for only 10-15% of the forms ( Table1 ). However, the patterns of phonetic variation are relatively straightforward to describe within a syllabic framework.... In PAGE 8: ... One of the hallmarks of human discourse is its flexible format and expression. There are many different ways to pronounce common words ( Table1 ) and embed these within a grammatical framework that is itself rather fluidly defined. Despite such variability in linguistic expression, listeners readily interpret the speech stream.... ..."
Table 3: Performance measures with and without derivatives
in Interval extensions of non-smooth functions for global optimization and nonlinear systems solvers
1996
"... In PAGE 13: ... (That is, the midpoint test was not used.) The results for these algorithm variants appear in Table3 . There, the subcolumns labeled IN denote the variant with the interval Newton meth- ods, those labelled MT denote the variant with the monotonicity test, but no interval Newton method,, and those labelled NONE represent the basic algorithm without the monotonicity test.... ..."
Cited by 5
Results 1 - 10
of
13,271
